A primary element of a portable computer system is a display. Since Cathode Ray Tube (CRT) displays are relatively large and heavy, with high power requirements, other alternatives have actively been sought. Flat panel display technology represents a significant alternative to CRT display technology. Flat panel displays may have several advantages over CRT displays. Flat panel displays include a number of different display types, Liquid Crystal Display (LCD) being most commonly used. LCD displays may have advantages of being compact and relatively flat, consuming little power, and in many cases displaying color.
Typical disadvantages of LCD displays may be poor contrast in bright light—especially bright natural light, inconsistent performance in cold temperatures, and display resolutions which may be constrained by a fixed number of row elements and column elements. Among these limitations, fixed resolution may cause significant problems for LCD operation in a multimedia environment. Multimedia users may demand a monitor which can be configured for different display resolutions. Analog CRT displays may be easily configured for different resolutions.
Flat panel displays may typically comprise two glass plates pressed together with active elements sandwiched between. High resolution flat panel displays use matrix addressing to activate pixels. Conductive strips for rows may be embedded on one side of a panel and similar strips for columns are located on the other side. Panels may be activated on a row by row basis in sequence. This process may be described in more detail in a text entitled: “High Resolution Graphics Display Systems”, Peddie 1994 (pp. 191–225), incorporated herein by reference, however the general nature of LCD addressing is known in the art.
LCD flat panel display resolution may be dictated by physical construction of an LCD. CRT displays have a continuous phosphor coating and may be illuminated by an analog signal driving an electron beam. Because of the analog nature of CRT, scaling display resolution is relatively simple. LCD displays have a fixed array of physical pixels which may be turned on or off by applying or removing a charge.
While resolution of a CRT may be changed by changing scanning frequency parameters, LCDs are limited by a fixed number of row and column elements. Fixed resolution LCD displays are particularly troublesome in multimedia systems. Such systems may require changes in display resolution to take full advantage of applications displaying high resolution graphics. In addition, for a manufacturer of display controllers to claim full VGA, SVGA, and XGA compatibility limitations of fixed panel resolution must be overcome.
TABLE 1Vertical scanning frequencies for different graphics display modesTypicalVertical ScanPanel TypeResolutionFrequencyVGA Panel640 × 48025 MhzSVGA Panel800 × 60040 MhzXGA Panel1024 × 768 65 Mhz
Like an analog CRT, an LCD panel may be controlled by a horizontal and vertical scanning signal. Data may be displayed in its respective screen position during an interval in time corresponding to when vertical and horizontal scan signals for a particular location coincide. Horizontal and vertical scan signals are set at a frequency proportional to display resolution. Table 1 contains vertical scanning frequencies for popular graphics display modes.
Typical vertical scanning frequencies may be 25 MHz for 640 pixels by 480 pixels display, 40 MHz for 800 pixels by 600 pixels, and 65 MHz for 1024 pixels by 768 pixels. New panels comprising 1280 pixels by 960 pixels may have an even higher vertical scanning frequency. A high resolution display therefore may have a higher scanning frequency than a relatively lower resolution display.
Using the general principal stating high frequency is proportional to high resolution, some downscaling may be achieved by attempting to replicate lower scanning frequencies of low resolution display while maintaining native scanning resolution. On a fixed resolution display of 600 pixels by 800 pixels for example, a 640 pixel by 480 pixel resolution output may be scaled by lowering the frequency at which data is clocked to the display. This type of approach to expansion related problems may be considered synchronous. Synchronous approaches may have disadvantages for expanding certain resolutions.
Because of the relationship between scan frequencies for certain resolutions that need to be expanded, synchronous approaches to expansion may not be desirable. Visual anomalies such as flicker, and related line dropping may cause noticeable and annoying visual artifacts. Also, horizontal flicker may be noticed and is even more annoying as portions of the display shift from side to side. This is due to the inability of the expansion scheme to account for every line generated at one resolution to a corresponding line on a second resolution. Resolutions which divide evenly into each other may be best suited for synchronous approaches.
Asynchronous approaches may be necessary when the ratio of CRT display lines and LCD display lines, based on different desired display resolution and fixed resolution display capability, is non-integral and when it is generally considered desirable to decouple the time base upon which display data is generated from the time base upon which output display resolution is generated. Consider an example when 3 LCD display lines must be displayed for every 2 CRT lines.
Prior art methods use relatively expensive dual path approaches which may replicate hardware for each display sought to be driven. In addition to hardware costs, bandwidth requirements may be approximately doubled and available bandwidth cut by approximately half for a dual path approach. Other disadvantages of a dual path approach may be non-transparency of software. With a dual path approach, display related software may require separate modification to standard register contents, standard addresses or the like in order to operate at each resolution.
For transforming graphics resolutions, fewer problems are inherent in downscaling, when desired display resolution is larger than the panel. Upscaling however may present special problems. When attempting to display lower resolution graphics on a higher resolution, fixed resolution panel display a variety of compensation methods may be used. Compensation features may be made available through use of shadow registers and extension registers. Both compensation method and desired parameters, such as output resolution may be set through use of registers.
Some systems employ a compensation technique known as centering. With centering, a smaller resolution graphic image may be placed within a larger resolution display in the center of the display. One problem associated with centering a 640 pixel by 480 pixel display at full color within, for example, a 1024 pixel by 768 pixel display is limited bandwidth. On a display which supports 640 pixels by 480 pixels in native mode (e.g. at native 640 pixel by 480 pixel timing of 25 Mhz), there may be sufficient bandwidth to support 24 or 32 bits per pixel of color.
As frequency increases such as on a fixed panel 1024 pixel by 768 pixel display which does not support the native timing for 640 pixels by 480 pixels resolution, bandwidth requirements increase in proportion to increase in frequency between resolutions. Most 32 or 64 bit controller may only support 24 or 32 bit full color at a native resolution of 640 pixels by 480 pixels. Another problem with centering and prior art expansion techniques is the scope of programming required to support it. Many shadow registers must be programmed, and protection mechanisms must be in place to configure and then preserve the expanded display settings.
FIG. 1 is a diagram illustrating a prior art technique of centering. During centering, Graphics Window 200 with a resolution of 640 pixels by 480 pixels may be displayed on Fixed Resolution Panel 201 which is capable of displaying at a fixed resolution of 1024 pixels by 768 pixels. Graphics Window 200 may be generated by a software application such as a computer game with high resolution graphics. For consistency and compatibility purposes, such a computer game may generate a display with a resolution of 640 pixels by 480 pixels regardless of the resolution capability of the display.
Differences in size must be accommodated to physically center a smaller display within a larger resolution panel. Additionally, differences in normal VGA timing which may be around 25 Mhz, and native timing of an LCD panel which, for a 1024 pixel by 768 pixel display, may be around 65 Mhz must be accommodated. In other words, during centering, a panel must actively accommodate the difference between lower resolution graphics mode and higher resolution panel by generating blank pixels. The resulting display is often too small to be viewed acceptably. For a 1024 by 768 pixel panel there may be 9 or 10 inches of display surface of which one third may go unused during centering. Not only does this waste panel capability, but refresh rates are poor because of timing translation and often the displayed information is too small to read either in Windows™ or in DOS text mode. From an economic standpoint, a user pays a premium for the increased resolution of the panel display only to receive inferior performance.
Another compensation technique for vertical scaling is known as line replication. In line replication or stretching, every Nth line may be duplicated on a subsequent line. In text mode, blank line insertion may be used to evenly fill an entire panel.
Yet another problem arises when attempting to drive two display devices with different display resolutions either through a SIMULSCAN™ output or an auxiliary output. For example, if Microsoft™ Windows™ is running, a dual display mode may be activated by way of an icon as is done for SIMULSCAN™ displays. Requests may then be passed by Windows™ Graphic Driver Interface (GDI) to an appropriate display driver and hardware. Only one graphics resolution, however, may be selected for one or both displays at one time. In other words, separate display resolutions may not be desirable for each display in a particular SIMULSCAN™ environment. Thus, on a notebook system with an 800 pixel by 600 pixel LCD display, if a 640 pixel by 480 pixel resolution is chosen, for example, to drive an external LCD projection panel as a SIMULSCAN™ output, then the LCD output must either be “centered” as described earlier or otherwise accommodated.
Typically, fixed resolution panels present the most difficulties in graphics scaling since other elements may more often be flexible. Every resolution capable of being generated by a system must be capable of being displayed on a fixed panel for true compatibility. Some CRT based projection systems, however, may be inflexible as to timing and resolution parameters and thus must be used in their native resolutions only. This native resolution may present special difficulties as it may use non-standard timing or resolution.
A typical native resolution for projection CRT displays is 640 pixels by 480 pixels. Use of fixed resolution projection systems leads to problems with fixed resolution panels in cases where projection system resolution does not match panel resolution. In such a case, shutting off LCD panel display may be an undesirable alternative. Another undesirable alternative may be the dual path method previously described which allows independent display of any two resolutions.
When such multimedia display equipment is used with conventional portable computers, because of fixed resolution related problems, a single display resolution only may be displayed on both displays (internal or projected) at the same time. In many instances, it may be desirable to project presentation material on an external monitor while displaying other information (e.g., speaker's notes) on an internal display. It may also be desirable to switch between internal and external displays, such that a speaker may preview an image prior to projection display. Furthermore, a need for two video displays containing different images may arise in other situations where computers are used, such as CAD systems, spreadsheets, and word processors. In particular, use of Windows™ may make it desirable to allow a user to open one window (or application) on a first video display (e.g., laptop flat panel display) and open another application on another display (e.g., external monitor). Thus, for example, a user may be able to display a scheduler (daily organizer) program on one display while operating a word processing program on another.
Popular prior art approaches to providing multiple displays with different images driven by one computer such as in the dual path method previously described have disadvantages beyond mere hardware cost. In lap-top or notebook computers, dual path methods may increase power drain, weight and size in addition to cost. Minimizing power, cost, size, and weight is especially critical in highly competitive notebook computer markets.
Other methods to drive two displays involves two display signals sharing refresh rates. To faithfully provide two distinct display resolutions, it may be desirable to generate two separate signals for two video displays having different resolutions, pixel depths, and/or refresh rates. For example, it may be desirable to generate two displays in different graphics modes, or one display in a graphics mode and another in text mode. Moreover, two different displays (e.g., flat panel display and CRT) may use refresh rates different from one another. Alternately, one display may provide improved performance operating at a particular refresh rate unavailable for the other display. In the context of upscaling an image to a fixed resolution display however, traditional methods such as interpolation may not be available or may be inefficient.
Interpolation is a well-known prior art technique used for upscaling video images. In an interpolation scheme, several adjacent pixels in a source video image are typically used to generate additional new pixels. During vertical interpolation of source image data, throughput performance problems may be encountered in a scan-line-dominant-order-of-storing scheme because vertical interpolation usually requires pixels from different scan lines. Accessing different scan lines may require retrieving data from different pages of display memory forcing a non-aligned or non-page mode read access. A non-page mode read access may require more clock cycles than a page mode access for memory locations within a pre-charged row. Thus average memory access time during vertical interpolation may be much higher than consecutive memory accesses within the same row. High average memory access time during vertical interpolation may result in a decrease in the overall throughput performance of a graphics controller chip.
To minimize number of accesses across different rows, a graphics controller chip may retrieve and store a previous scan line in a local memory element. For example, with respect to FIG. 2, a graphics controller chip may retrieve and store all pixels corresponding to scan line A-B and store retrieved pixels in a local memory located in a graphics controller chip. The graphics controller chip may then retrieve pixels corresponding to scan line C-D, and interpolate using pixels stored in local memory.